v EVOLUTION AND TELEOLOGY 345 



A-B-C, etc. Such a sequence, though involving changes, 

 may be uniform in the sense that given A, its change into B 

 and C regularly follows, and it may be possible to show 

 inductively that this sequence depends on no external con- 

 ditions. If so, A ' as such ' yields B and C. Further, we 

 may be able to c explain ' this sequence if we can analyse A 

 into elements which, either severally or in their action upon 

 another, can be viewed as a process of transition towards B. 

 It is, in fact, only when we can thus regard change as con- 

 tinuous that we can be said to understand it. Explanation 

 of events in series, then, resolves itself into the search for 

 continuity in permanence or continuity in change. 



But, secondly, the problem of change may be of quite 

 another kind. Comparing two cases that are partly similar 

 two cases, say, of A we find them to be alike in one 

 respect and different in others there, for instance, as AB, 

 here as AD. How are we to reduce variation of this kind 

 to uniformity? We begin, as a rule, with a search for 

 antecedents. A is explained as the effect of a, B of /5, and 

 so on. But it is clear that we do not, on these lines, resolve 

 the variable relation AB into a uniform relation, i.e. one 

 that holds between the terms as such. We may find an 

 antecedent a which, as such, gives rise to A, and an ante- 

 cedent ft which, as such, gives rise to B. But it will be 

 only the relation a/3 which gives rise to AB. Now if the 

 relation a/3 holds between the terms as such, it is uniform, 

 and it follows that the relation AB is uniform also. But 

 AB varies, and it was its variation which we had to resolve 

 into uniformity. It is clear that we cannot succeed on 

 these lines. The relation aft must also be variable. If 

 we ask, in turn, for the uniform relation which is to explain 

 it, and look for it in the antecedents of a and ft severally 

 the same argument will repeat itself and so ad infinitum. 



A variable collocation AB, then, has antecedents aft 

 or ab, which also form a variable collocation. We shall 

 not, by tracing the antecedents further, find a relation hold- 

 ing between the terms as such. There must be a condition 

 of the collocation outside the related terms as such, and 

 outside the series of their causes as such. Let us call it C. 

 Then the position must be that given C we have AB, 



